05 02-33-27PoplackLeveyInPress

language and Space

An International Handbook of Linguistic Variation

Volume 1: Theories and Methods

Edited by

Peter Auer and JUrgen Erich Schmidt

Offprint

De Gruyter Mouton

IV. Structure and dynamics across language spaces

22. Co ntact-induced g rammatica l cha n ge: A ca utiona ry tale

I. Inlroduction ') Background 3. Establishing change 4. Applying the comparative variationist framework to the investigation of change 5 . Putt ing contact-induced change to the test 6. Discussion 7. Acknowledgement 8. References

1. Introduction

When the space between two languages - or their speakers - is reduced enough to result in contact, a variety of outcomes may ensue. Most prominent by far is change. So ubiqu i tous has the association between contact and change become that i t is invoked to account for a wide array of developments that differ from some (typically ill-defined) norm. This is the case even where the candidate development already exists in the recipient language, where no other-language form is visible, and where the transition period is as short as a single generation. And despite the general consensus that grammatical structure is particularly resistant to transfer, reports of dramatic structural changes i n languages known t o have been i n contact are legion.

In thi s article we suggest that much of the cvidence brought to bear on contacti nduced change - diachronic as well as synchronic - either fails to demonstrate that change has occurred, and/or if it has, that i t is the product of contact and not i nternal evolution. These issues, together with the possibil i ty that the inherent variability characteristic of all spoken language may have been mistaken for change, need to be resolved before a contact explanation can be justified.

In the ensuing sections we f irst provide a brief overview of some of the more salient types of change reported to result from contact . We then review the kinds of confounding evidence that plagues the study of change, contact-induced or otherwise, and propose criteria for establishing its existence empirically. Tn section 3 we detail a method, combining the machinery of variationist sociolinguistics with the comparative method of historical l inguistics, capable of identifying change and ascertaining its source. We then present a series of case studies illustrating the ut il ity of this approach and conclude with some observations about the implications of our findings for understanding the existence. nature and extent of contact-induced change. Our focus is on the grammar, but thc considerations we present also apply to other typcs of change, with the exception of lexical borrowmg.

Poplack & Levey 2010

IV. Structure and dynamics across language spaces

2. Background

Contact-induced structural change has long been a subject of debate among linguists, due in no small part to the primacy traditionally accorded the internal cvolution of language (e.g., Lass 1980; Martinet 1960). This was traditionally assumed to be regular and rule-governed, while external ly-motivated changes could be arbitrary, unpredictable and idiosyncratic. External cxplanations were thus often invoked as a last resort, whcn a more satisfactory account was not available (Romaine 1988: 349).

Early opinions ranged from outright denial of the possibility of a mixed language (e.g., Mi.iller 1871: 86; Whitney 1881) to the elaim that there are no completely unmixed languages (Schuchardt 1884: 5). But most of those who admitted the possibility of contact-induced change did so cautiously and under specific conditions. Haugen's comment that "[on] those relatively rare occasions when bound morphemes are actual ly transferred from one language to another, they arc such as fit readily into a preexistent category in the recipient language" (1954: 386), is but one example. Similarly, lakobson ( 1962: 241) argued that "a language accepts foreign structural elements only when they correspond to its own developmental tendencies". Notwithstanding such caveats, major struetural ehanges said to result from language contact continue to be reported.

Among the more spectacular changes of the past figure the borrowing of Turkish inOectional suffixes into dialects of Asia Minor Greek ( Dawkins 1916), the incorporation of the entire Spanish preposition system into Malinche Mexicano (Field 2005: 352), the borrowing of Russian finite verb morphology into Mednyj Aleut (Thomason 2001), the development of Michif from French noun phrases and Cree verb phrases (e.g., Bakker 1997; Bakker and Papen 1997) and the (in)famous convergence of Kupwar Urdu, Marathi and Kannada into "'a single surface syntaetic strueture" (Gumperz and Wilson 1971: 155) .

After reviewing many sueh ehanges, a s well a s a variety of theories o f what constrains them, Thomason was led to reject the idea that eategorical constraints against contactindueed change could be enunciated. Instead she proposed that under the right social and linguistic conditions,

[ . . . J anything goes, i ncluding structural borrowing that results in major typological changes in the borr owing language. In phonology, loss or attrit ion of entire phone tic and/or phonological categories in native words and of all kinds of morphophonemic r ules. In syntax, sweeping changes in such features as word order, relative clauses, negation, coordination, subordination, comparison, and quantification. I n morphology, typological ly disruptive changes such as the replacement of flexional by agglutinative morphology or vice-versa, the addition or loss of morphological categories that do not match in source and borrowing languages. and the wholesale loss or addition of agreement patterns. (Thomason 200 I: 71)

At the time of this writing, Thomason's (and Schuchardt's) views enjoy wide support. The implicit assumption is that change is an almost inevitable result of language contact (e.g., Appel and Muysken 1987: 154; Bynon 1977: 240; Harris and Campbell 1995: 149,

among many others; although see Thomason 2007 for discussion of deliberate resistance to contact-induced change). To be sure, some scholars continue to caution against promiscuous inferences, particularly in the absence of robust sociohistorical and linguistic evidence (e.g., Dorian 1993; Mougeon, Nadasdi and Rehner 2005). And Heine and Kutcva (2005: 8) observe that contact is orten invoked without satisfactorily ruling out


22. Contact-induced grammatical change

other causal mechanisms. Nonetheless, the onus seems to be on the analyst to justify why change did not occu r in an apparently favorable situation:

But since [the above] predictions are robust - that is. they are valid in the great majority of cases that h ave been described in the l i terature - any rio/a/ioll should provide i nteresting i nsights into social , and. to a lesser extent . l inguislic determinants of contact- induced change. (Thomason 200 I: 71. emphasis ours)

And in fact, the kinds of changes enumerated above are reported in contemporary contact situations as wel l . In the least conspicuous cases, an existing minor-use pattern is elaimed to be developing into a major-use pattern via overextension of recipient-language options into new contexts (e.g., Heine and Kuteva 2005: 44-62; K lein 1 980; SilvaCorvalcln 1 994a: 1 67; Toribio 2004: 1 67). Thus Toribio (2004: 1 70) states that when the "bilingual mode" of two Spanish-English bilinguals was experimentally induced, their use of overt subject pronouns, while not ungrammatical in Spanish, reflected the inf1uence of English, a non-pro-drop language. Savic ( 1 995: 487, cited in Heine and Kuteva 2005: 69) reaches similar conelusions in relation to slightly elevated uses of optional overt subject and object NPs in the discourse of Serbian-English bilinguals . Heine and Kuteva also discuss an increase of overt pronouns among Russian and H ungarian bilinguals, again on the model of English (2005 : 68), as do Otheguy, Zentella and Livert (2007) for New York City Spanish and Backus (2005: 333) for Turkish. The samc proccss h as been said to initiate word order changes, as in the shift from the Spanish majority pattern of postposed adjectival placement (fa persona mits importal1te) to the prc-nominal adjectival p lacement (fa 111(1.1' il71portal1te persol1a) characteristic of English (Toribio 2004: 1 67).

In another type of change, recipient-language morphemes lllay encode a completely new grammatical category, as in Tariana-Portuguese bilinguals' use of interrogative pronouns (e.g., k1l'(111({ 'who') to mark relative and complement elauses on the model of Portuguese, or of Portuguese lexemes to express the Tariana evidentiality marking system (Aikhenvald 2002: 1 82 - 1 83, 3 1 5 - 3 1 6) .

Other contemporary cases of convergence involve outright violation of recipicnt-Ianguagc grammar. Clyne (2003: 1 32) discusses overgeneralization by Dutch-English and, to a lesser extent, German-English bilinguals of SVO word order in contexts where SOY and VSO are prescribed in the non-contact varietics.

Apart from extcnsion of an existing option, addition of a totally new category and violation of recipient-language rules, contact is widely believed to accelerate attrition and loss of grammatical features in one language with no equivalent in the other (e.g., Seliger and Vago 1 99 1 ). The subjunctive mood in New World Romance varieties is frequently cited as an example of contact-induced erosion (see e.g., Garcia 1 985; Gutierrez 1 994; Lantolf 1 978; Ocampo 1 990; Silva-Corval�ll1 1 994b for Spanish and L aurier 1 989 for Canadian French). We explore this phenomenon in more detail below. Other morphosyntactic elements said to be receding as a result of contact include the BE auxiliary (sl'il1. :ijn. essl'fl') in Australian German, Dutch and ftalian respectively, in favor of haben. heMen, opere 'have' . on the modcl of English, which admits only hm'e in such contexts (Clyne 2003: 1 18- 1 1 9) . Clyne does acknowledge regional variation in auxiliary usage, even in some non-contact varicties of these languages, but follows Thomason and Kaufman ( 1 9811: 511- 59), in arguing that contact reinforces language-internal tendencies. And these examples co uld be multiplied ad IihiflllJl.

393


394 IV. Structure and dynamics across language spaces

Much of the evidence for contact-induced change may lend itself equally well to alternate explanations. Sometimes inherent variability is equated with change (as in the aforementioned Australian example) . Elsewhere, perceived differences from a standard arc designated changes (as in the increase of overt subject pronouns). And it is often unclear whether the change in question is idiosyncratic or community-wide. The next section summarizes these and other pitfalls and details some of the considerations that must be taken into account in the empirical establishment of change.

3. Establishing change

3.1. Variability is not change

The hallmark of speech is inherent variability, alternate ways of saying the same thing (Labov 1969). Variability is a necessary condition for change, but is not, in and of itself, coterminous with it. In many areas of the grammar, alternations among variant forms may persist for centuries, but linguists who believe that language is invariant often interpret them as signs of change. This inference is particularly prevalent when the forms in question are detected among speakers or groups considered vulnerable to external influence (e.g., bilinguals, residents of minority-language communities in intense contact with a majority language, etc.). Change is also routinely invoked when one of the variants is salient or stigmatized , especially when a null form is involved. Thus, the widespread variability between subjunctive and indicative morphology in (prescribed) subjunctiveselecting contexts cited above is usually characterized in contact situations as loss of the subjunctive, despite the fact that this alternation has been attested for centuries (Poplack 2003; St-Amand 2002). So a first requirement is to determine whether variability is involved in change. Variant choice is never random, but is instead highly constrained by features of the (linguistic and extra-linguistic) environment. Measuring the trajectory of variant use over time, coupled with analysis of the fine conditioning of variant choice, provide a sensitive and refined diagnostic of change, in ways we illustrate in section 5 below.

3.2. Locating an earlier stage

A t the most trivial level, change shows up as a dillerence (qualitative or quantitative) between an outcome and an earlier stage, but in much of the current literature, the structure of the earlier stage is either not appealed to, or at best, only inferred (e.g., Winford 2003: 64). Absent systematic comparison with a relevant precursor, any claims about what has changed must remain speculative. (Indeed, lack of an appropriate diachronic baseline is at least partially responsible for the widespread assumption that many salient features of contemporary vernaculars, both contact and non-contact, are recent innovations; see also Zwicky 2005).

The first step in establishing the existence of change is comparison over time. This may not be simple or s traightforward, given the orten fragmentary nature of surviving diachronic evidence. This abiding land at times insoluble) problem in historical linguis-



tics has givcn rise to an entire methodology devoted to reconstruction (e.g., Campbell 1999, 2003; Hoenigswald 1960; Lehman 1992; Ringe 2003, to name but a few). Synchronically, however, with a modicum of ingenuity, earlier stages or surrogates for them can orten be found.

3.3. The nature of the reference point

Not every earlier stage is equally useful in assessing change, however. In most reported cases, the comparison point (diachronic or synchronic) is some poorly defined "standaEf'. Constituted only of forms that have been prescriptively ratified, the standard rarely acknowledges variants that cannot be associated (rightly or wrongly) with variant functions or readings (Poplack et al. 2002). As a result, the language ratified as standard may diverge considerably from the language actually spoken, which orten develops independently of normative injunction (Poplack and Dion to appear; Leroux 2003; Poplack and Malvar 2007; Willis 2000). The variability most characteristic of vernacular speech involves competition among standard and nonstandard features. When compared with an idealized invariant standard, the latter may appear to be changes.

A nice illustration of the importance of an appropriate reference, or baseline, variety comes from a recent study of Spanish in contact with English in New York City (Otheguy, Zentel1a and Livert 2007). As noted above in connection with the extension of minor-use patterns, rates of overt pronouns are thought to be increasing among secondgeneration Spanish speakers, on the model of English, in which the subject is assumed to be categorically expressed (but see Leroux and Jarmasz [2006] for conditions governing variable subject pro-drop in English) . Of course, some overt pronoun expression is tolerated in Spanish, but the standard baseline does not specify how much. What then constitutes an "increase"? Vis-cl-vis what? Otheguy, Zentel1a and Livert's comparison with one surrogate pre-contact stage of Spanish (spoken by newcomers to New York) revealed, contrary to the widely espoused convergence hypothesis enunciated above, an even higher rate of overt pronouns than that displayed by their New York born-andraised counterparts (2007: 786). Only once the appropriate comparison variety was identified ( i . e. by distinguishing two newcomer dialects, each with a different initial rate of overt pronouns), were they able to demonstrate that pronoun expression had in fact increased in the second generation.

Invidious comparisons can only be avoided if the reference variety is commensurate with the variety hosting the candidate for change. If it is a spoken vernacular, the baseline should also be a spoken vernacular (or an appropriate surrogate thereof ); if it is a contact variety, the baseline should be a pre-contact variety, and so on. As we will show below, when the inference of change is pursued scientifically via systematic comparisons (vertical and horizontal) of appropriate reference varieties, potential candidates orten turn out either not to be changes, or not to be contact-induced, but rather cases of garden-variety inherent variability firmly rooted in the internal structure of language (Leroux and Jarmasz 2006; Poplack et a!. 2006). Efforts invested in locating and mining appropriate earlier stages have paid off handsomely (Ayres-Bennett 2000; Gordon et a!. 2004; Martineau and Mougeon 2003; Poplack and St-Amand 2007). The procedure is a good deal more laborious than simply inferring what the precursor was like, but in view of the potential for misidentification, we see this as a requirement for any claim of change bascd on synchronic data.

395


396 [V. Structure and dynamIcs across language spaces

3.4. The social embedding of change

Another major impediment to detecting change resides in the nature, quantity and quality of the evidence. Manifestations of change are relatively rare in running discourse. This may be why much of the published evidence comes from data collected in experimental or quasi-experimental conditions (e.g .. informant elicitation, subjective reaction tests, even "forced" interviews with semi-speakers [see. e.g. , Altenberg 1991; HuJlines

1991; Kaufman and Aronoff 1991; Montrul 2002; Schmidt 1991]) . Atypical situations may entail atypical behavior, with l ittle or no bearing on what transpires in real-life bilingual settings (Poplack 1983: 116). Without knowing how well, if at all, these experiments rellect language use in the community, their results cannot be interpreted.

This brings us to the distinction between innovatiun and change. Every change involves an innovation in the speech of one or more individuals and its subsequent diffusion across the rest of the community (Croft 2000; Joseph and Janda 2003: 13; Milroy 1993: 223; Shapiro 1995: 105, fn. 1; Thomason 2007). For some l inguists, the innovation is the change, and it is assumed to occur during first language acquisition. Theoretically, we would argue (along with Labov 1994: 3 10-3 1 1 and most other sociolinguists), that "an innovation in a speaker's output is not a linguistic change until it has been agreed

on and adopted by some community of speakers" (Milroy 1992: 221), af()l·tiori because the vast majority of changes do not "catch on" (e.g., Thomason 2001: 130) . From an empirical perspective, however, this is a moot point: in all but the rarest of cases, the moment of innovation is simply unobservable.

We can rephrase the issue in terms of the relationship between the individual and the community. Claims of change are often based on exiguous speaker samples (e.g., Toribio's 2004 study of two individuals), if any. The linguistic behavior of the individual may be ephemeral, situation-bound and/or simply deviant from the norms of the remainder of the community (Labov 1972a; Poplack 1983: 1 19). especiaJly when this behavior is elicited under laboratory conditions. Alternatively, it may be a predictable outcome of one or more explanatory factors, linguistic and/or extra-linguistic. Only once the speaker's linguistic behavior is situated with respect to that of the appropriate reference variety (other speakers of the same age, bil ingual ability, educational level, etc.) is it possible to decide. If change has occurred, it should be observable (at least to some degree) in the speech of the relevant members of the community (e.g., the younger cohort, if spearheaded by youth; the highly bil ingual, if induced by contact, etc . ) . In other words, in the absence of diffusion. change cannot be confirmed.

3.5. Type of bilingual population

The locus of the candidate for contact-induced change is also important. As Bul lock and Toribio ( 2004: 9 1) observe. researchers have drawn on an array of bi lingual populations.

ineluding heritage language speakers, incomplete acquirers and those undergoing lan

guage loss. Shifting popUlations have garnered most attention (see, e.g., Dorian 1981, 1989: Gal 1979; Mougeon and Beniak 1991: Otheguy. Zentella and Liven 2007; Si lvaCorvalill1 1991), but the linguistic behavior they display is not necessarily applicable to stable bi lingual communities (Poplack 1 993: Winford 2003: 64). Tn the widely studied


22. Contact-ll1duced grammatical change

contexts of language restriction and obsolescence, structural differences (from the mainstream or non-contact variety) may in fact be inevitable, given that many of the informants may not (or no longer) use it with any regularity. In etTect, then, although ethnically minority-group members, they may be L2 speakers of the minority language (Poplaek 1997: 3(5). Lz speakers can expected to make all manner of acquisition errors, depending on their level of proficiency at the time of data collection. Though these may be transitory, in contact situations, they too are often construed as changes.

3.6. The linguistic embedding of change

In addition to a certain level of diffusion across the community, the innovation should also occur (to some degree) in relevant linguistic contexts. Relevant contexts are determined by situating the innovation in the larger host linguistic system-synchronic and diachronic. Among the questions to be answered are: To what extent has the innovation gained a foothold? What is its current role? Has it replaced a native form in one or more functions? Did it introduce a new function into recipient-language grammar? Isolated, anecdotal or exceptional examples, so often cited in the literature, reveal nothing about the regularity of the innovation, its productivity, or the extent to which it is entrenched in the language (Weinreich, Labov and Herzog 1968: 185). This can only come from studying the sustained discourse of speakers in context. Because changes examined synchronically are likely to still be in progress, and thus continue to feature variability, these questions can only be addressed by establishing the structure of the variability, as it emerges from the constraints conditioning variant choice.

3.7. Recognizing change

Most reported changes are tangible: a novel lexical element (e.g., a borrowing), unwarranted presence of an existing form (e.g. , increased pronoun expression in pro-drop languages), absence of a prescribed form (e.g .. the French subjunctive), or choice of the (prescriptively) "wrong" form (e.g., auxiliary Iwben for seil1 in Australian German or moir for Clre in Canadian French) . But observed changes tend to be far more subtle. especially during the sometimes very lengthy transition period from one language state to another. Linguists who focus on inherent variability recognize, in ascending order of importance, change in frequency of one or more competing forms, change in statistical significance of one or more factors contributing to variant choice and change in linguistic structure. With rare exceptions, these elude casual observation, but can be detected with the aid of the methods described in section 4.

3.8. Distinguishing contact-induced change from internal evolution

Once it has been established that a change has inueed taken place, it remains to demonstrate that it is contact-induced and not the product of drift. This stcp is usually skipped in the literaturc. much of which is predicatcd on the aforementioned (widespread but

397


I V. S tructure and dynamics across language spaces

unfounded) assumption that linguistic differences occurring in bilingual contexts are necessarily 1) changes and 2) contact-induced. Thomason (2007: 42) defines contact-induced change as follows: "a particular linguistic change is caused at least in part by language (or dialect) contact if it would have been less likely to occur outside a particular contact situation". Given how little we know about the likelihood of particular changes - within or outside contact situations - we offer alternative criteria:

A candidate for contact-induced change in a contact variety is presen t in the presumed source variety and either I) absent in thc pre-contact or non-contact variety, or 2) if present (e.g. , through i nteri ingual coincidence), is not conditioned in the same way as in the source, ({nd 3) can also be shown (0 parallel in some non-trivial way (he behavior of a counterpart feature in the source.

Compliance with these conditions can only be ascertained through systematic quantitativc comparisons, of a diagnostic linguistic feature, with an earlier or pre-contact stage (or a surrogate thereof), with a non-contact variety, and most important, with the presumed source. As we shall see in what follows, these criteria have not been met by most extant studies, with the result that the existence, frequency and extent of contact-induced change remain unclear.

Summarizing, sufficient understanding of the outcomes of contact appears beyond the reach of anything but systematic corpus-based research carried out within the bil ingual community. Application of an empirically accountable quantitative methodology to large and representative bodies of bil ingual speech can contribute significantly to resolving the difficulties associated with ascertaining whether contact-induced change has authentically occurred. It is to such a methodological framework that we now turn.

4. Applying the comparative variationist framework to the investigation of change

4.1. The variationist paradigm

The approach that we exemplify in the remainder of this article draws on the combined resources of variationist sociolinguistics (e.g., Labov 1972a, Labov 1972b; Labov et al. 1968; Weinreich, Labov and Herzog 1968) and the Comparative Method of historical linguistics (e.g., Baldi 1990; Campbell 1998; Hoenigswald 1960; Meillet 1967) . A central tenet of the variationist paradigm is that the recurrent choices made by speakers in discourse, though not completely predictable, are not random, but are subject to constraints imposed by "the phonological cnvironment, the syntactic context, the discursive function of the utterance, topic, style. situation and personal and/or sociodemographic characteristics of the speaker or other participants" (Sankoff 1988a: 15 1) . Thc kcy theoretical construct of this framework is the linguistic variable (Labov 196611982), which comprises a set of mriants among which speakers alternate in exprcssing a given meaning or function. This results in a heterogeneous but structured linguistic systcm. A foundational working principle is that the structure - grammatical and social - underlying the heterogencity can be inferred from the distribution and conditioning of competing variants in discourse.


22. C:ontact-lI1duced grammatIca l change

4.2. Communities and speakers

We noted above that change proceeds from diffusion. and extent of diffusion can only be assessed in reference to a community of speakers. A crucial step. though orten neglected. is to locate an appropriate (here, bilingual) communit)' in which to carry out the study. The nature and makeup of the community allow us to test hypotheses about the extra-linguistic factors often claimed to be relevant to contact-induced change. These include intcnsity (if contact (the more intense. the greater the likelihood of structural interference). Icngth or contact (the greater the time-depth. the greater the likelihood of change). status of the languages in the community (minority or majority) and si:::c of speaker population (the fewer and more marginal the speakers. the more likely that they will borrow from the language of the dominant group) (e.g .. Romaine 1995; Thomason 200 I; Winford 2(03) . Study sites can be selected to represent these parameters and to test whether these factors are or are not operating.

Speakers constituting these communities can be sampled to represent individual dimensions of the contact axis. Sampling protocols are also guided by hypotheses about how contact-induced change works. According to one. individual levcl of bilingualism is inversely correlated with mastery of the minority-language grammar, such that the greater the proficiency in the other language. the more likely the possibility of simplification or loss of minority-language linguistic structure. Another proposes that the incorporation on the local level of majority-language material (via code-switching or borrowing) into otherwise minority-language discourse may bring with it associated grammatical properties. and these in turn may lead to structural convergence (e.g .. Backus 2005; King 2005; Muysken 2000; Thomason 200 1; Winford 2(03). According to this hypothesis. a speaker's individual propensity to code-switch would be a predictor of contact-induced change. Individual attitudes toward. and relative prestige of. each of the languages could also affect the extent and direction of change (e.g .• Appel and Muysken 1987: 158; Joseph 2007; Romaine 1995: 66; Thomason 2(0 1) .

The case studies in section 5 demonstrate how these and other hypotheses can be tested on linguistic production data provided by speakers representing different combinations of these conditions, thus enabling us to ascertain which are most explanatory of contact-induced change and which community members arc most likely to lead it.

4.3. Sampling bilingual speech

Once the appropriate speakers have been identified. we can record samples of their spontaneous speech (bilingual ({nd monolingual) in quantities sufficient to permit accountable analysis. The speech of choice for such an endeavor is the venwculw: Not only does the vernacular offer "the most systematic data for linguistic analysis" ( Labov 1984: 29). it is also the register in which linguistic manifestations of language contact are most likely to occur. Variationists mine vernacular corpora to uncover the regular patterns that characterize natural exchanges in the speech community and to distinguish them from isolated or idiosyncratic occurrences. A pattern may be defined as "a series of parallel occurrences (established according to structural and/or functional criteria) occur-


40() IV. Structure and dynamics across language spaces

ring at a non-negligible rate of use" (Poplack and Meechan 1998: 129). Patterns can only be discerned by systematic quantitative analysis of the data in accordance with the principle of ac("olilltahility (Labov 1972b : 72) . This requires not only that all the relevant occurrences of a feature be incorporated into the analysis, but additionally all the contexts in which it could have occurred but did not. Adherence to the principle of accountability allows the analyst to identify anomalous occurrences which arc neither representative nor recurrent (and thus unlikely to trigger change). while identifying regular tendencies, which /Jill)" (or may not) herald change.

4.4. Variable rule analysis

Certain linguistic and extra-linguistic conditions can be expected to promote change, though none is fully determinant. Claims that even favorable linguistic factors can be overridden by the right constellation of social factors (e.g., Thomason 200 l : 77; Winford 2003 : 25) raise the issue of the relative weights each contributes to a particular outcome and the relationship of such weights to different types of contact situations. This is a problem that lends itself well to multivariate analysis, the major analytical tool in the variationist framework. Variable rule anal)'sis ( Rand and Sankoff 1990; SankotT, Tagliamonte and Smith 2005) facilitates extraction of regularities from the welter of data making up running discourse. The multiple regression procedure incorporated within it helps determine which aspects of the linguistic (e.g., phonological, morphological, syntactic, etc.) and extra-linguistic (e.g., minority status, bilingual ability, age, etc.) contexts contribute statistically significant effects to variant choice when all are considered simultaneously (e.g., Sankoff 1988b). These aspects are hYjJotheses about the structure of variant choice. Variable rule analysis enables us to operationalize them (as factors) and determine which give the best account of the data. The goals are to ascertain whether change has taken place and if so, whether it is the product of contact or internal development.

Much empirical work on contact-induced change in progress revolves around frcquencies. As detailed by Poplack and Tagliamontc (200 I: 92), however, differences in rates of variant selection must be used with caution to infer change, contact-induced or otherwise. The key diagnostic in assessing thc relationship and provenance of forms is the constraint hierarchy, the configuration of environmental factors affecting the probability that a given variant form will be selected, along with the direction of their effccts. We construe these as a portion of the grammar underlying the variability. To the cxtent that constraint hierarchies are language-specific. they are powerful indicators of change (see scction 5).

Similarities in surface form in contact and source varieties can be misleading. Thcy may result from borrowing or transfer. justifying the inference of changc. but may also be due to interlingual coincidence or to linguistic universals. To rule out altcrnative explanations, we rely on con/licf sites (Poplack and Meechan 1998: J 32): functional, structural and/or quantitative difli'rences, typically manifested as a contlict in constraint hierarchies. The conflict site. together with detailed cross-variety comparison. plays a

crucial role in detecting change and identifying its ultimate provenance. If the hicrarchy of constraints conditioning the variable occurrence of a candidate for change (e.g .. 0

subjunctive. ovcrt subjcct pronoun. preposition stranding, null relative marker. perinhrastic future) in a contact varietv is the same as that of its pre-contact precursor, while



dilfering from that of its presumed source, no structural changc has taken place. If it features a constraint hierarchy different from those of both its prc-contact precursor ({nd the prcsumcd source, we can infer that changc has occurred. but not one that is contactinduced. Only when a candidate for change in a contact varicty fcaturcs a constraint hierarchy diffC'/"{:'lIf from that of its pre-contact precursor, but p({j"{{lIcl to that of its presumcd source. can we concludc in favor of contact-induccd change.

4.5. Cross-variety comparison

Comparisons are at the heart of inferring changc, and as such, figure, at least implicitly, in all studics of contact-induced change. But many havc been carried out without reference to the principle of accountability, or involve an idealized comparison variety. Still others are simply anecdotal. Innovative in the research program we describc hcre is the extent of the scientific apparatus brought to bear on the comparison. The comparative variationist enterprise (Poplack and Meechan 1998; Poplack and Tagliamonte 2001; Tagliamonte 2002) contributes not only a methodological aspect - the construction, statistical analysis and linguistic interpretation of suitable comparison varieties (pre-contact, non-contact and post-contact), but also the critical capacity to decide among competing hypotheses. The focus is on identifying empirical criteria capable of both detecting change and ascertaining its source, and testing hypotheses empirically to determine the goodness of fit with the data. This involves principled data collection, enunciation and operationalization of hypotheses and their statistical evaluation in large-scale corpora. In section 5, we illustrate this approach by considering a number of case studies which attest to the versatility of quantitative variationist methodology for identifying contactinduced change.

5. Putting contact-induced change to the test

The studies we present below were carried out in a context of long-term intense contact bctween French and English in Canada, a situation that meets all the criteria considered t�lvorable to structural convergence (e.g., Thomason and Kaufmann 1988; Willford 2003: 90). Depending on locality, each of the languages has a different olTicial status: English is the majority language ill Ontario, while in Quebec, French assumes this role. This configuration allows for comparison of each of the languages in minority and majority guisc. The 284 speakers involved in these studies arc distributed among these four conditions. Other characteristics relevant to the contact situation include density of contact on the local levcl (as expressed by the ratio of anglopbones to francophones in the community). individual bilingual ability (as measured by a bilingual proficiency index [Poplack 1989, Poplack, Walker and Ma1colmson 2006]) and relative propensity to borrow or code-switch (as inferred by comparing individual rates of each across sample members) . Standard extra-linguistic l�lctors like social class, education. etc. were also taken into account. The inference or change ovcr time was investigated by comparing the specch or younger and oldcr generations.

401



Thc linguistic data come from two massive corpora of spontaneous speech collected from these speakers: the OIIC/ll'a-Hull French Corpus (Poplack 1989) and the Quebec English Corpus (Poplack, Walker and Ma1colmson 2006). The typological similarity of French and English should render them particularly hospitable to contact-induced change. In addition, superficially similar structures were deliberately targeted, as these are considered to be prime candidates for grammatical convergence. Indeed, at least two of the cases described below had been reported to result from this process.

5.1. Attrition of the subjunctive in Ottawa-Hull French

In the most straightforward case, we locate, in community speech, a linguistic feature that is used differently from the way it is prescribed and determine which individuals are deviating the most. If the difference is the result of change, and if that change is contact-induced, its leaders should figure among the younger, the more bilingual and those in most intense contact with the majority language. The French subjunctive is an excellent candidate. It is (prescriptively) obligatory after the class of "subjunctive-selecting" main verbs, like fal/oir in example (I), but in Canadian French it has been found to be variable «(1); Poplack 1990) . (This and all ensuing examples are reproduced verbatim from audio recordings from one of the corpora described above. The codes in parentheses refer to corpus [QEC: Corpus of Quebec English; OR: Corpus of Ottml'a-Hull French], speaker and line number. The examples cited throughout this paper are instantiations of patterns that emerged from the large-scale quantitative analyses presented in sections 5.1- 5.3)

(I) F(dlclit qu' elle repol/{/ (IND) "oui, tu pellx faire trois pas de Necessary (PAST) that she answer (IND) yes you can make three steps of

gcant ".

giants

F{dlail qu' elle repollde (SUBJ) la phrase Necessary (PAST) that she answer (SUBJ) the sentence

'She had to say "yes, you may take three giant steps". sentence. '

complete. complete.

She had to say the whole (OH.025.2186)

Because the subjunctive is no longer used productively in English, the (variable) occurrence of the indicative in contcxts like ( I ) is inferred to be a stage in the loss of this form, a development which Laurier (1989) characterized as the result of contact.

Table 22.1 displays the results of a multivariate analysis of the extralinguistic factors contributing to choice of subjunctive. In this and the ensuing tables, the higher the figure. the greater the probability the "application" variant will be selected in the environment listed on the left. Table 22. I reveals a small, but statistically significant, tendency for the most bilingual speakers to use least subjunctive. This result supports the hypothesis that high levels of bilingualism lead to linguistic simplification and loss. Most studies would stop at this demonstration (if they provided any quantitative information at all). But if this is a contact-induced change, effects of the other factors should point in the same dircction. This not the case. Instead, the two French-majority-Ianguage communities in Quebec. Vicux Hull and Mont Blcu, which should favor the subjunctive most under this



Tab. 22.1: Variable rule analysis of the contribution of extralinguistic factors to the choice of sub

junctive under verbal matrices

Corrected mean Total N

SEX

AGE

Female

Male

45-54

55-64

15-24

25-- 34

65+

35-44

NEIGHBOURHOOD OF RESIDENCE

Vieux Hull, Quebec (FR majority working class)

West End, Ontario (FR minority working class)

Vanier, Ontario (FR minority upper-working class)

Basse-Ville, Ontario (FR minority working class)

Mont Bleu, Quebec (FR majority upper-middle class)

ENGLISH PROFICIENCY

mid-low

low

mid-high

high

0.715

2694

.52

.46

.55

.53

.52

.50

.48

.44

.57

.51

.50

.50

.43

.54

.53

.48 ·····;42

hypothesis, instead use most and least, depending on social class. (Working-class speakers, even more counter-intuitively, apparently use it the most.) Also militating against the inference of change is the fact that the age distribution is not monotonic: the youngest speakers are as likely to select the subjunctive as their oldest counterparts. Further analysis revealed that these apparent extra-linguistic effects are in fact epiphenomena of a strong lexical effect (Poplack 1990, 1992: 2001). Subjunctive morphology is highly associated with only a few main verbs, and because those verbs happened to be unevenly distributed across the subgroups represented in Table 22.1, this showed up as a discrep-

Tab. 22.2: Variable rule analysis of the contribution of English proficiency to the choice of subjunc

tive in verbs embedded under four classes of verbal matrix

English Proficiency

low mid-low mid-high high

VERBS EMBEDDED UNDER

high frequency/high subjunctive matrix verb:

fct/luir .55 .58 .40 .42

high frequency/high subjunctive matrix verbs:

vouloir / aimer .46 .58 .42 .60

low frequency/low subjunctive matrix verbs .68 .47 .49 .31

low frequency/variable subjunctive matrix verbs .34 .60 .61 .39

403



Tab. 22.3: Variable rule analysis of the contribution of socioeconomic class to the choice of subjunc

tive in verbs embedded under four classes of verbal matrix

Socioeconomic class

unskilled skilled sales / profes-

service sional

VERBS E MBEDDED UNDER

high frequency/high sUbjunctive matrix verb:

.lid/oir .45 .53 .47 .69

high frequency/high sUbjunctive matrix verbs:

vIIII/oir / (limer .36 .62 .64 .52

low frequency/low subjunctive matrix verbs .56 .47 .42 .58

low frequency/variable sUbjunctive matrix

verbs .50 .42 .42 .68

ancy in subjunctive usage. When the lexical identity of the main verb is factored out, as in Table 22.2, bilingual proficiency no longer shows a systematic effect.

The real explanation for differential use of the sUbjunctive is class-based, and is not relevant to the contact situation at all, as illustrated in Table 22.3 (see Poplack 1997 for detail). This result could not have been intuited from casual comparisons with either the idealized standard benchmark (which prescribes the subjunctive categorically), or amongst proficiency groups. This is but one illustration of our earlier contention that differences in overall rates of variant occurrence need not be indicative of change, contact-induced or otherwise. Indeed, real-time analysis over a time span of 120 years (StAmand and Poplack 2002) showed that the current situation of the subjunctive is actually a retention.

5.2. Preposition stranding in Ottawa-Hull French

It has been argued that a local factor, the extent to which individual speakers draw on the two languages in discourse, is a better predictor of contact-induced change than macro-level factors like minority status or density of contact. Under this hypothesis, speakers who code-switch the most would lead contact-induced change (Backus 2005; Bullock and Toribio 2004; King 2005; Thomason 2001; Winford 2003). Zentz (2006) was the first to test this hypothesis empirically, using as a diagnostic the syntactic phenomenon of preposition stranding in Canadian French, as in (2).

(2) COl11l11e Ie gars que je sors aI'ee, lui il' parle- bien if est jral1r,:ais

Like the guy that I go out with him he speaks- well he IS French 'Like the guy I'm going out with, he speaks- well he is French.' (OH.013.2122)

Because preposition stranding is normatively proscribed, examples like (2) have been ascribed to contact, if not convergence, with English (e.g., King 2000; Roberge and Rosen 1999), in which stranding is presumed to occur freelv.



Tab. 22.4: Independent variable rule analyses of the contribution of l inguistic factors to preposi tion orphaning and stranding for low and high code-switchers ( brackets ind icate that the factor was not statistically signi fican t )

Corrected lIIe(//1 Total N

SYNTACTIC FACTORS Place of topic

Discourse referent In tra-sentential referent

Type of construction Cleft Object precedes relative Pseudo-cleft

Presence of another complement

None Additional Post-V complement Cli tic

SEMANTIC FACTORS Semantic contcnt of preposition

"Weak" "Strong"

Obligatoriness of prepo-sition

Obligatory Non-obligatory

Humanness of noun complement

Non-human Human

Orphaning

"Low" Code-switchers

.09 982

N/A

N/A N/A

0

.42

.61

.57

.36

"High�' Code-switchcrs

.02 662

N/A

N/A N/A

.19

.96

Stranding

"Low" Codc-switchers

.Of 224

N/A

N/A

.35

.40

.99

.19

.99

.46

.82

"High" Code-switchcrs

JJ3 116

N/A

N/A

.19

.95

in the contact language with what is preslimed to be the case in the source. Absent any accountable study of preposition stranding in the contact variety of English, we cannot be fully confident about exactly what was transferred. In additional, there is no evidence (in this or other studies), that the current state of affairs represents a change. Demonstration of contact-induced change would minimally include a vertical comparison (preferably with a pre-contact stage, but at the very least with an earlier stage), a horizontal comparison with a non-contact varietv and comparison with the structure of the source.



5.3. Relative marker expression in Quebec English

A study by Poplack et al. (2006) illustrates the unexpccted results to emerge from such an approach. It involves the variable use of overt and null relative markers in accusative relative clauses in Quebec English, a minority language in intense contact with French. This is illustrated in example (4). Quebec French shows the same (surface) alternation in the same context (S).

(4) a. I said" Oh yeah, I remember III'as lI'earing lhis disgustingl)' ugfy green suit that you made me ll'ear", or something. (QEC.03I.II3S)

b. Ill'as always kn01l'n as Carrot Head and Freckle Face and a fell' other l'er)' pulgar names @ they called me. (QEC.006.602)

(S) a. Bien jt comprenais fe .fi'wu;ais, mais c' esl les mots que je comprenais

Well 1 understood the French but it is the words that 1 understood pas.

not 'Well, T understood the French, but it's the words that I didn't understand.'

(OH.OS9.812)

b. Les maisol1s @ ils faisaient dans ce temps hI fe bois ctail Ires-The houses 0 they made 111 that time there the wood was very-if Crait pas cher pal1toUie.

it was not expensive at all 'The houses they madc in those days, the wood was very- it wasn't expensive at all.' (OH.OSO.698)

In particular, both languages permit a null variant (4b, Sb), although this is prescriptively unacceptable in French. How is it generated? Table 22.S displays the factors constraining choice of null variant across mainstream varieties and cohorts of contact-variety speakers.

A first important observation is that the structure of variable relative marker expression differs in the two non-contact varieties. In English (column 2), the strongest predictor of a null relative marker is the presence of intervening material between the subject of the relative clause and its antecedent; in French (column I), the major effect is phonological. Because these effects are language-specific, we can use them as diagnostics of contact-induced change. We may now compare the contact variety to its non-contact counterpart (columns 2 and 3).

For all factor groups, the constraint hierarchies are parallel. This means that, in the aggregate, the contact variety of English is no different from its mainstream countcrpart. But according to the hypotheses enunciated earlier, certain subgroups of the popUlation could be expected to spearhead changes. So we will want to establish whether all of the minority-language speakers behave in the same f�lshion, by comparing the conditioning of variant choice according to speaker age and bilingual ability. Predictably, the behavior or the older (column S) and least bilingual (column 7) speakers is very similar to that of speakers of noncontact English (column 2) in terms of constraint hierarchies. But the younger (column 4) and more bilingual sDcakers (column 6) share less with the main-

407


4()� I V. Structurc and dynamics across language spaccs

Tab. 22. S : Independent variable rule analyses of the contribution of l inguistic factors to t he selection of the IlII// relative marker: across mainstream varieties

Corrcc/ed 1111'(l/l:

Tot a l N

I n tervening material

None

Present

Anteccdent type

Pronolln

NOLIn

Type or subject or the

rel a t i ve clause

Pronoun

NOLIn

Following phonologi

cal environment

Sonorants

Obstruents

Defi n i tcness of ante

cedent

Delinite

Indefinite

Sentence type

Cleft

Other

Mainstream

French

.ON SOl460

. 7 1

.44

o

. 1 3

.94

. 6 1

.44

2 3

Mainstream Contact

English English

.367 . 53] 1 09/288 I SOl28 I

4

Younger

speakers

..:too I -, I n 1 6

Oldcr

speakers

.50] 1 �S/253

. 5 3

. 1 8

.60

. 3 1

stream in terms of the fine condit ioning of variant choice. This kind of result i s consistent with their predicted role as leaders of change. Are these in fact changes? Comparison with older speakers suggests that they are. But are they contact-induced? As we noted earlier, to sustain any elaim in this respect, the involvement of the source language (here. French) mLlst be established. The three major "departures" of the younger and more bilingual cohorts with respect to their older and less bilingual counterparts involve the effects of antecedent type, type of subject of the relative and following phonological environment. The first two elTeets have been neutralized in the younger generation. while in French, as shown in column 1, they are strong and significant. The effect most relevant to the contact situation is phonological. Recall that this is the major predictor of variant choice in French, and the younger and most bilingual anglo phones are the only ones who display i t . This result is consistent with the inference of contact-induced change. But comparison of constraint hierarchies shows that the French effect goes in the opposite direction. (The apparent English phonological effect is an epiphenomenon of a developing tendency for certain pronominal subjects which arc coincidentally obstruent- init ial (he, she, Ihey) to disfavor deletion.) Here again, di fferences among minori ty-language cohorts, suggestive as they appear on the surface, cannot be attributed to convergence



Tab. 22.6: Independent variable rule analyses of the contribution of l inguistic factors to the selec

tion of the l1uff relative marker: younger contact vs. benchmark speakers

Cortl'e/I'd 1111'(/11:

Total N

Intervening material

None

Present

Antecedent type

Pronoun

Noun

Type of subject of the relative clause

Pronoun

Noun

Fol lowing phonological environment

Sonorants

Obstruents

Definiteness of antecedent

Definite

Indefinite

Younger contact

speakers

. 482 831 1 68

. 5 6

. 1 8

[ .53]

[. 38]

. 59

.40

Younger mainstream

speakers

.299 481 1 48

. 5 5

. 1 0

.76

.45

[.54]

[ .34]

with French. To what then are they due? Comparison with the non-contact benchmark provides the answer (Table 22.6). When speakers of mainstream English are analyzed according to age, we find the same apparent changes in progress.

In this case our method has enabled us to detect change - the young minoritylanguage anglophones are diverging from their elders (column 5 of Table 22.5) - and to attribute it to the correct source: participation in a change in progress in mainstream English.

These examples, along with many others (e.g., Dion 2003; Jarmasz 2005; Leroux and Jarmasz 2006; Petrik 2005; Poplack and St-Amand 2007; St-Amand 2002) demonstrate that when the inference of contact-induced change is pursued systematically, it becomes increasingly difficult to justify. Particularly instructive is the fact that even in quantitative studies, differences between cohorts or overall rates may be masking effects that are independent of the contact situation.

6. Discussion

Change is widely considered a predictable, if not inevitable, outcome of language contact. The l iterature is rife with accounts of radical alterations of the grammars of contact languages, whether by virtue of incorporating forms or functions with no antecedent in the recipient language, or losing one or more of its core grammatical elements. Many of these changes would appear to be very rapid, occurring even in the space of a single

40Y


4 1 0 I V. Structure and dynamics across language spaces

generation. This is consistent with another widespread. if unproven. idca: that contact accelerates change. On both of these counts. contact-induced change appears suspiciously difTerent from internal evolution. which we know to be gradual. moderate and vastly more conservative.

Observing that scientific delllonstration of change, contact-induced or other. was often missing from this research. we reviewed some of the p henomena that could lead the analyst down the garden path of inferring change, even where such an inference is not warranted. Chief among them are the inherent variability characteristic of spontaneous speech and the dearth of accountable corpus-based studies of the languages in contact. W hen confronted with an idealized standard variety. but lacking systematic comparison with the more relevant pre- and non-contact varieties. or with the supposed source variety of the claimed change. we detailed how the (variable) occurrence of non-standard variants could easily be confused with change.

We also outlined a number of conditions necessary to establish the existence of contact-induced change. Many had already been enunciated elsewhere (Poplack 1993 ; Poplack and Meechan 1998); some were recently reiterated (e.g., Backus 2005: 3 10). The following are paraphrased from Thomason (2001: 93-94):

I. Situate the proposed change with respect to its host linguistic system 2. Identify a presumed source of the c hange 3. Locate structural features shared by the source and recipient languages 4. P rove that the proposed i nterference features were not present in the pre-contact

variety 5. Prove that the proposed interference featu res were present in the source variety prior

to contact 6. Rule out (or situate) internal motivations

Thomason herself concedes that " in many. possibly even most contact situations around the world we cannot at present satisfy requirements 4 and 5". going so far as to caution that "if we can't satisfy all the requirements, then we can't make a solid case for contactinduced change" (Thomason 200 1 : 94). This raises the issue, which has preoccupied us throughout this article, of just how many of the contact-induced changes currently treated as facts satisfy any of these conditions, let alone all. This is a very important question, since these "facts" arc the basis for our theories about the nature, extent and mechanisms of contact-induced change.

We observed earlier that for many diachronic changes, the crucial evidence would now be i mpossible to reconstruct. But as we have been at pains to illustrate, this is certainly not the case for synchronic situations. It is thus all the more puzzling that the requisite information is missing from all but a handful of contemporary studies. Why should this be? The growing. but patently false. perception among l inguists working in other paradigms that it is too difficult, i f not i mpossible, to access the relevant evidence

may play a role. The treatment of frequency is a case in point. F requency is at the root of most inferences of change, whether manifested as raw rates of occurrence of a form. or its diffusion across a community or linguistic system. Among the few who have addressed th is issue at all, a certain skepticism about the feasibility of ascertain ing freq uency empirically is detectible. l1ms with reference to the appearance of foreign elements i n bilingual speakers' discourse, Thomason observes, "determining the frequency of appearance is a hopeless task" (200 1 : 134). and goes on to q uery: "how can one


22. Contact-induced grammatlcai change

determine whether a speaker used [an element] once, occasionally or frequently?" Likewise, while Backus (2005) admits that most changes are l ikely of the "frequency-changing kind", he contends that these are precisely the changes that "are virtually impossible to demonstrate empirically" (2005 : 334). Such assertions betray a surprising lack of famil iarity with corpus-based studies, and fly i n the face of the important results of largcscale quantitative analyses that have been performed on languages in contact.

Despite recent concerns about the absence of an empirically well-founded framework for investigating bi lingual speech (Van Coetsem 2000: 39 - 40; Winford 2007: 22) and the alleged limitations of variation-based techniques for illuminating contact phenomena (Muysken 2000: 250), we submit that the rigorous comparative methodology presented here is capable of detecting change and distinguishing external from internal motivations. it thus discriminates among compet ing hypotheses about the nature of these processes.

Agreeing with Thomason (200 1 : 88) that the crucial similarity bet ween the processes of contact-induced and internal change has to do with competitio/1 (i. e. variability) between an innovative feature and an older feature, we have outl ined a procedure for assessing contact-induced change within a framework that takes account of its symbiotic relationship with l inguistic variation. We hope to have shown that reliance on the structure of variant choice affords a depth of analysis that transcends impressionistic observations based on perceived similar ities between surface forms and rates of occurrence.

Indeed, the very strength of the variation-based approach resides in its capacity to discern even minor perturbations i n constrain t hierarchies, which /11(/), be h arbingers of change, contact-induced or otherwise. Otheguy, Livert and Zentella (2007) provide the most recent confirmation that i t can detect even the smallest adjustments in constraint ranking across generations of bil ingual speakers.

Illustrating with studies of three well-documented candidates, we found no evidence of change in the first two, and the change documented i n the th ird was not contact

induced. Moreover, in l ine with Thomason's (200 1 : 9 1 -92) caveat that a "case for contact origin can't be made on the basis of single feature", additional research situating a wide variety of Canadian French and English vernacular variants i n historical, social and l inguistic context corroborates the findings presented here: none of the contemporary non-standard variants typically ascribed to either l inguistic change or language contact are in fact innovations (Poplack and D ion to appear; Elsig and Pop1ack 2006; Jarmasz 2005; Leroux 2007; Poplack 1993). Rather, virtually all were not only present in the pre-contact variety, but, more important, they were largely constrained in the same way.

T hese results are especially striking when we consider that all the popularly invoked prerequisites for contact-induced change - social (protracted contact, extensive bilingualism; high levels of proficiency in the respective source languages, etc.) and l inguistic (i. e. typological congruence [Thomason 200 I: 71], in terl i ngual equivalence [Backus 2005: 326]; grammatical "weak points" [Bullock and Toribio 2004: 91], attractiveness [Johanson 2002: 2]) - are met.

Why then did contact-induced change fail to occur? Given the present state of our knowledge, we can only speculate. T he length of contact may have been too brief to result in change, although a time span of two centuries is a good deal longer than those on which Illany recent claims have been based. Perhaps the morpho-syntactic features we examined are particularly resistant to change, although they were specifically selected on the basis of typological similarities thought to facil itate it. We cannot rule out the possibi l i ty that our analytical tools may have misidentified or excluded changes; we

4 1 1


4 1 2 I V. Structure and dynamics across language spaccs

reserve judgment until a demonstration to that effect is offered. Certainly the circumstances - linguistic, social and attitudinal - could not have been more propitious. But while favorable circumstances may encourage change, we know that they do not determine it. A remaining possibility is that contact-induced change is a good deal less common than the l iterature suggests. This is our best guess at the moment.

Of course, the analyses we have presented, despi te being based on corpora of bilingual speech with sizes ranging into the millions of words and large representative samples of individuals in contact situations, cannot begin to do justice to the full trajectory of change over time and across all speakers of a language. But opting for the alternative -anecdotal observation - simply because "it is much easier to show a qualitative change" (Backus 2005: 334) - ignores the important scientific advances in empi rical linguistics made over the last half century. Contact-induced change is not an inevitable, nor possibly even a common, outcome of language contact. Only more accountable analyses of more contact situations will tell. In the interim, the burden of proof is on those who claim that it has occurred.

7. Acknowledge ment

The research reported here was generously f unded by grants to Poplack from the Social Sciences and Humanities Research Council of Canada and the Killam Foundation. Poplack holds the Canada Research Chair in Linguistics at the University of Ottawa.

8. References

Aikhenvald, Alexandra 2002 Language conlacl in Ama�ol1ia. New York: Oxford University Press.

Altenberg, Evelyn P. 1 99 1 Assessing first language vulnerability to attrit ion. I n : Seliger a nd Vago (cds . ) , 1 89 - 206.

Appel, Rene and Picter Muysken 1 987 Languagc Conlllcl IIlld Bilingualism. London: Edward Arnold.

Ayres-Bennett, Wendy 2000 Voices from the past: Sources of spoken seventeen th-century French . Romanischl'

Forscil/lI1gcll 1 1 2(3) : 32R - 348. Backus. Ad

2005 Codeswitching and language change: One thing leads t o another') International JOl/rn1l1 of' Bilingualism 9(3/4): 307 - 340.

Baldi , Phi l ip 1 990 I n troduction: The Comparative Method. I n : Phi l ip Baldi (ed. ) , Linguislic Chonge an"

Recollstructioll I'vlelhodolog)", 1 - 1 3 . Berl in : Mouton de Gruyter. B a kker. Peter

1 997 "/I Language of" Ollr 011'11 " : The Genesis of" Michi/: the ivlixcd Cree-Frl'llch Lunguogl' of Ihe C{//wdion MClis. (Oxford Studies i n Anthropological Linguistics 1 0 . ) Oxford: Oxford University Press.

Bakker, Peter and Robert A. Papen 1 997 Michif: A mixed language based on Cree and French. I n : Sarah G. Thomason (ed . ),

COll/ac/ L(/ngllagl's, 295- 363. Amsterdam: J ohn Beniamins.


22. Contact- i n d uced grammatical change

Bullock. Barbara E. and Almeida Jacqueline Toribio 2004 In troduction: Convergence as an emergent property in bi l ingual speech. Bilingua/ism:

L({Ilguagl' (lnd Cognition 7(2) : 9 1 - 94. Bynon. Theodora

1 97 7 Histori("([/ Linguistics. Cambridge: Cam bridge University Press. Campbell. Lyle

1 998 liistoriCII/ Linguistics: An Introductiol1. Edinburgh: Edinburgh University Press. Campbel l , Lyle

1 999 Historica/ Linguistics: An Introduction. Cambridge. MA: M IT Press. Campbell . Lyle

2003 How to show languages a re related: Methods for distant genetic relationship. I n : Joseph and Janda (eds.). 262 -282.

Clyne. Michael 2003 Dynamics oj" Language Contact. Cambridge: Cambridge University Press.

Croft. Wil l iam 2000 Explaining Language Change: An El'O/utionarl' Approach. Harlow/New York: Longman.

Dawkins. Richard MacGil l ivray 1 9 1 6 Modern Greek in Asia Minor. Cambridge: Cambridge University Press.

Dion. Nathal ie 2003 L'elTacement du (/Ul' en fran<;:ais canadien: Une etude en temps reel. Unpublished MA

thesis (111ezl 10 ire). Department of Linguistics, University of Ottawa. Dorian, Nancy C.

1 98 1 Language Dmth: The Lij"e Cye/e o/"a Sculfish Gae/ic Dia/cct. Phi ladelphia, PA: University of Pennsylvan ia Press.

Dorian , Nancy C. (cd . ) 1 989 111 1'estigating Obsolescence: Studies in Language Contraction and Dmth. Cambridge: Cam

bridge University Press. Dorian, Nancy

1 993 Internally and externally motivated change in language contact settings: Doubts about dichotomy. I n : C harles Jones (ed.) , Jiistorical Linguistics. Problems and Per.ljil'ctil'es, 1 3 1 - 1 54. London: Longman.

Elsig, Marti n and Shana Poplack 2006 Transplanted d ialects and language change: Question formation in Quebec. Unil'l'rsity oJ"

Penll.lyll'a1lia Working Papas in Linguistics 1 0(2 ) : 77 - 90. [Selected Papers FUI11 N WA V 34. ]

Field, Frederic 2005 Long-term effects of CS: C lues to structural borrowing. In/fTlwtioIW/ Juurnal ul Bilingua/

iSI1 1 9(3/4): 341 - 360. Gal , Susan

1 979 Language Shi/i: Social Determinants oj"L inguistic Change ill Bilingual Austria. New York: Academic Press.

Garcia, Erica C . 1 985 Shift ing variation. Lingua 67: 1 89 - 224.

Gordon, El izabeth M. , Lyle Campbell, Jennifer Hay, Margaret Maclagan, Andrea Sudbury and Peter Trudgil l

2004 Nell' Zea/and English: Its Origins and El'olutioll. Cambridge: Cambridge University Press. Grevisse, Maurice

1 980 Le BOil Usage. Duculot : Paris. Gumperz, John J . and Robert Wilson

1 97 1 Convergence and creol ization: A case from the I ndo-Aryan/Dravidian border. In : Dell Hymes (eel . ) , Pidgil1 i::utiol1 and Creoli::ation oj" Lallguages, 1 5 1 l 67 . Cambridge: Cambridge University Press.

4 1 3


4 1 4 I V. Structure and dynamics across language spaces

Gutierrez, Manucl 1 994 Simplification, transfer a nd convergence in Chicano Spanish. The Bilingllal RCl'iclI'/ La

Rt'l'ista Bilingiie: 1 9 : 1 1 1 - 1 2 1 . H arris. Alice C. and Lyle Campbell

1 995 Historical Syl1 la.Y in Cross-Lingllistic Per.ljJectil'c. Cambridge: Cambridge Un iversity Press.

H augen, Einar 1 954 Review of Wei nreich ( 1 953 ) . Langllage 30: 380- 388.

Heine. Bernd and Tania Kuteva 2005 Langllage Contact and Grammatical Change. Cambridge: Cambridge University Press.

H oenigswald, Henry 1 960 Language Change and Linguistic Reconstruction. Chicago: Un iversity of Chicago Press.

H uffines. Marion Lois 1 99 1 Pennsylvania German: Convergence and change as strategies o f discourse. I n : Seliger and

Vago (eds. ) . 1 25 - 1 37. Jakobson, Roman

1 962 Sur la theorie des affin i tes phonologiques entre des Iangues. In: Stephen Rudy (ed . ) . Roman jako/Jsol1: Selcctcd J·jlritil1gs, vol. I , 2 3 4 -246. The Hague: Mouton.

Jarmasz. Lidia-Gabriela 2005 "1 th ink 1 know what I'm thinking": The alternation between simple and progressive

forms in stative verbs. U npublished M A thesis (n1l!l11oire), Department of Linguistics. U niversity of Ottawa.

J ohanson, Lars 2002 Structural Factors in Turkic Languagc Contact. London: Curzon .

Joseph, Br ian D . 2007 Attitudes and ideology in contact-induced change: Trumping all psychological and struc

tural factors. Paper presented at the 1 8th International Conference on H istorical Linguistics ( ICIi L) . Un iversitc du Quebec it Montreal, August.

Joseph, Brian D. and Richard D. Janda (eds . ) 2003 The Handbook of Historical Linguistics. Oxford : Blackwell .

Kaufman. Dorit and Mark Aronoff 1 99 1 Morphological disintegration and reconstruction. In: Seliger and Vago (eds.) , 1 75 - 1 88.

King. Ruth 2000 The Lexical Basis of Grammatical Borroll 'ing. Philadelphia/Amsterdam: John Benjamins.

K ing. Ruth 2005 Crossing grammatical borders: Tracing thc path of contact-induced l i nguistic change. In :

Markku Filppula. Juhan i Klemola. M,lljatta Palander and Esa Pen tiIHi (cds . ). Dialects Across Borders: Selected Popel's .limn the /! th Internatiollal Con/erence Oil !vlethods in Dialectology ( Methods XI) , jO('llSUU, August 2002. 233 - 25 1 . Amsterdam/Phi ladelphia : .Iohn Benjamins.

K lein. Flora 1 980 A quantitative study of syntactic and pragmatic indications of change in the Spanish of

bil inguals in the U .S. In : Wil l iam Laboy (ed.) . Locating Language ill Tilllc and Space. 69 - 82. New York: Academic Press.

Labov. Will iam [ 1 966] 1 982 The Social Strati/lmtion u{ Englislz in Nell ' York City. Washington: Center for Ap

plied Linguistics. Laboy, Wil l iam

1 969 Contraction. deletion and i nherent variabi l i ty of the English copula. Language 45(4): 7 1 5 - 762.

Labov. Wil l iam 1 972a LIIIH!uaf!c in 'hc II111Cr Cit l'. Philadelohia: U niversitv of Pennsvlvania P ress.


22. Contact-induced grammat ica l change

Labov, Wil l iam 1 972b Sociolinguistic Pal/ems. Philadelphia: University of Pennsylvania Press.

Labov. Will iam 1 984 Field methods of the project on language change and variation . In : John Baugh and Joel

Schezer (eds . ) , Language in Use, 2 8 - 5 3 . Englewood Cliffs, N.I: Prentice Ha l l . Labov, Will iam

1 994 Principles of" Linguistic Challge: Intemal Factors. Oxford: Blackwell. Labov, William. Paul Cohen, Clarence Robins and John Lewis

1 968 A study of the non-standard English of Negro and Puerto Riean speakers i n New York City. Final report, Co-operative Research Report 3288, vol. I , US Regional Survey.

Lantolf, James 1 978 The variable constraints on mood in Puerto Rican Spanish. I n : Margarita Sui'icr (cd . ) ,

Contemporary Studies ill Romance Linguistics, 1 93 - 2 1 7 . Washington, D C : Georgetown University Press.

Lass. Roger 1 980 On E.yplaining Language Change. Cambridge: Cambridge University Press. Laurier, Michel

1 989 Le subjonetif dans Ie parler fi'anco-ontarien: Un mode en voie de d isparition') I n: Raymon Mougeon and Edouard Beniak (cds. ), L I' Fran("ais Canadien Parle Hors Quebec: AJiI'l"('u Sociolil1guistique, 1 05 - 1 26. Quebec: Presses de I ' Universite Laval .

Lehmann, Winfred P. 1 992 Historical L inguistics. 3rd cd. London/New York: Routledge.

Leroux, Martine 2003 Le passe recompose: Analyse diachronique et mul tivariee de la valeur temporelle du passe

en fran<;ais oral de Hu l l (Quebec). Unpublished MA thesis (memoire), Department of L i nguistics, University of Ottawa.

Leroux, Mart ine 2007 Something old, something new, something borrowed, something true: What are nul l sub

jects in French? Paper presented at New Ways of Analyzing Variation 36, University of Pennsylvania, Philadelphia, October.

Leroux, Martine and Lidia-Gabriela Jarmasz 2006 A study about nothing: Null subjects as a diagnostic of convergence between English and

French. Unil'ersity of" Peml.lyll'(/I1ia Working Papers ill Linguistics 1 2(2 ) : 1 - 1 4. Martineau, France and Raymond Mougeon

2003 A sociol inguistic study of the origins of )11' deletion i n European and Quebec French . Language 79( 1 ): 1 1 8 - 1 52.

Mart inet . Andre 1 960 £lh)1enl.\· de Linguistique Generale. Paris : Colin.

MeiJ let. Antoine 1 967 771(' Comparatil'e Method in Historical Linguistics. Paris: Librairic H onore Champion.

Mi l roy, .lames 1 992 Linguistic Variation und Chunge: On the Hislorical Sociolinguistics ()f" English. Oxford:

Blackwel l . Milroy, James

1 993 On the social origins of language change. In : Charles Jones (cd . ), Historical Linguistics: Prohil'ms und Pl'r.ljJl'ctires, 2 1 5 - 236. London: Longman.

MontruL Silvana 2002 Incomplete acquisit ion and attrition of Spanish tense/aspect distinctions in adult bi l in

guals. Bilingualism: Language and Cognition 5: 39-68. Mougeon, Raymond and Edouard Beniak

/ 99 1 Linguistic Consequences oj" Language Contacl and Rl'striction: The case oj" French in Onlario. Oxford: Oxford U niversitv Press.

4 1 5


416 I V. Structure and d ynamics across language spaces

Mougeon. Raymond. Terry Nadasdi and Katherine Rel1l1er 2005 Contact-ind uced l inguistic i nnovations on the continuum of language usc: The case of

French in Ontario. Bilingualism: Language and Cognition 8 (2 ) : 99- 1 1 5 . M ii l ler. Max

1 87 1 Lectures on the Sci('l7cc oj" Language. 1 st ed. New York: Scribner Muyskcn. Pieter

2000 Bilingual Speech: A Typology oj" Cude- /I!/i.Ying. Cambridge: Cambridge U n iversity Press. Ocampo. Francisco

1 990 The pragmatics of word order in constructions with a verb and a subject. Hispllnic Linguistics 4( 1 ): 87- 1 27.

Otheguy. Ricardo. Ana Celia Zentella and David Liver[ 2007 Language and dialect contact in Spanish in New York: Towards the formation of a speech

community. Language 83: 1 - 33. Petrik. Katrina

2005 Deontic modality in Quebec English: " Everything you need to know". U npubl ished M A thesis (mhl1oirt:). Department of Linguistics. University of Ottawa.

Poplack. Shana 1 983 B i l ingual competence: L inguistic interference or grammatical integrity? Tn : Lucia Elias

Olivares (cd. ) . Spanish in the Us. Selling: Beyond the Soutll 1 l "est. 1 07 ·- 1 3 l . Arlington: National Clearinghouse for Bi l ingual Education.

PopJack. Shana 1 989 The care and handling of a mega-corpus. Tn: Ralph Fasold and Deborah Sehi ffri n (cds . ) .

Language Change and Variation. 4 1 1 -45 1 . Amsterdam: Benjamins. Poplack. Shana

1 990 Prescription. intuition et usage: Ie subjonctif fran�ais et la variabi l i tc inhcren te. Langage et Societe 54: 5 - 33.

Poplack. Shana 1 992 The inherent variability of the French SUbj unctive. In: Christiane Laufer and TerrcH Mor

gan (cds. ) , Theoretical Analyses in Romance Linguistics. 235- 263. Amsterdam: Benja-1l11l1S.

Poplack. Shana 1 993 Variation theory and language contact. I n : Dennis Preston (ed . ) American Dialect RC'

smrch: An Anthology Celebrating the 100t17 Annil'ersmT oj" the AlIlerimn Dialect Society. 25 1 - 286. Amsterdam: Benjamins.

Poplack, Shana 1 997 The sociolinguistic dynamics of apparent convergence. Tn : Gregory Guy. John Baugh

and Deborah Schi tTri n (eels . ) . TrJ l l·ar{/.l· a Social Scienc(' oj" Lmzgllage. 285- :l09. Amsterdam: Benjamins.

Poplack. Shana 200 1 Variability. frequency and productivity in the i rrealis domain of French. I n : Joan Bybee

and Paul H opper (cds. ) , Fn'l[lIe1lcy ({nd the Emergenc(' oj" Linguistic Structure. 405 -428. Amsterdam: Benjamins.

Poplack, Shana 2003 Language contact and the evolution of French. Paper presented at Linguistic Symposium

on the Romance Languages ( LSR L) 33 and Conference on French in North America. I ndiana Un iversity. Bloomington, Apri l .

Poplack. Shana and Nathalie Dion to appear " Prescription versus praxis: The evolution of future temporal reference i n French'" .

Language. Poplack. Shana. Nathal ie Dian, Lidia-Gabriela Jarmasz, Carmen LeBlanc and Nicole Rosen

2002 Repertoire llistorique des Gmllllllaires <iu Francais. Corpus and documentation . Sociol inguistics Laboratorv. Un iversitv of Ottawa .


22. Contact-in d uced grammatical change

Poplaek. Shana. Adrienne Jones. Al l ison V. Lealess. Martine Leroux. Chclsea T. Smith. Yukiko Yoshizumi. Lauren Zentz and Nathal ie Dion

2006 Assessing convergence in contact languages. Papcr presented at New Ways of Analyzing Variation 35. Thc Ohio State University. Columbus. Novembcr.

Poplack. Shana and El isabctc MalvaI' 2007 Elucidating the transition period in l inguistic change. Probus 1 9( I): 1 2 1 - J 69.

PopJack . Shana and Marjory Meeehan 1 998 H ow languages fi t togcther i n code-mixing. Special issue of the International Journ{[1 of"

Bilillgualis/JI 2(2) : II/stant Loans. Ea.IT Conditiol/s: the Producti)'ity oj" Bilingual Borr(J l l "illg: 1 27 - 1 38 .

Poplack. Shana a n d Anne St-Amand 2007 A real-time window on 1 9th century vernacular French: The Rr'cits du ji'{[n(,ais (jllr'hi:cois

d'autreji)is. L{[I/guagl' in Socif!ly 36(5) : 707 - 734. Poplack. Shana and Sali Tagliamontc

200 I Aji-ican AlIIeric{{/z Ellglislz in the Diaspora. Oxford: Blackwell Publishers. Poplack. Shana. James Wal ker and Rebecca Malcolmson

2006 An English " l ike no other"'): Language contact and change in Quebec. Can{[dial/ Journal oj" Lillguistics. 5 1 (2/3) : 1 85 -2 1 3 .

Rand. David and David Sanko!T 1 990 Gold1'({rb 2. 1: A )'ariah/e rule application ./i)1· thl' J\1acilllosh. Montreal. Centre dc Re

cherches Mathematiques, University of Montreal . Version 2. ht tp://www.crm. ul110ntreal. ca/�sankoff/GoldVarb_Eng.html.

R inge. Don 2003 I nternal reconstruction. In : Brian D. Joseph and Richard D. Janda (eds . ). The Handbook

oj' Historical Linguistics, 244- 26 1 . Oxford: Blackwel l . Roberge, Yves and Nicole Rosen

1 999 Preposit ion stranding and que-deletion in varieties of North-American French. Linguistics AtlwHim 2 1 : 1 53 - 1 68 .

Romaine. Suzanne 1 988 Pidgin and Creole Lmzguag('s. London: Longman .

Romaine, Suzanne 1 995 Bilingualism. 2nd cd. Malden, MA: Blackwell.

Sankoff. David 1 988a Sociolinguistics and syntactic variation. In: Frederick J . Newmeyer (ed.) . Linguistics: Tilt!

Cambridge Surre l', vol . 4. Language: The Socio- Cultural Conte.Yt. 1 40 - 1 6 1 . Cambridge: Cambridge University Press.

SankolT. David 1 988b Variable rules. In: U lrich Ammon. Norbert Dittmar and Klaus J . Mattheier (cds. ) . Socio

linguistics: An Intemational Handbook ol tlze Sciencl' oj" Language and Societl '. vol . I . 984 - 997. ( Handbtichcr zur Sprach- und Komunikationswissensehaft 3 . 1 . ) Berl in : de Gruyter.

SankofL David. Sal i Tagliamonte and Eric Smith 2005 Guldl'arb X: A Variable Rule Application /in' iviacintosh ({nd WindOl l's. Toronto: Depart

ment of Linguistics. University of Toronto. Savie, Jelena M.

1 995 Structural convergence and language change: Evidence from Serbian/English codcswitching. Language in Socil'ty 24(4): 475 - 492.

Schmidt. Annette 1 99 J Language at t rition in Boumaa Fijian and Dyirbal . I n : Seliger and Vago (eds.) . 1 1 2 - 1 24.

Schuchardt. H ugo 1 884 Sh/l\ '(}-deulcizes 1I11d Slal l'o-italiclliscizes. Graz: Leuschner und Lubenskv.

4 1 7


4 1 � TV. Structure and dynamics across language spaces

Seliger, Herbert and Robert M . Vago (eds.) 1 99 1 First Languagl! Allritiol1. Cambridge: Cambridge University Press.

Shapiro. Michael 1 995 A case of distant assimilat ion: Istrl - , I ftr/. Aml!l'ican Speech 70: 1 0 1 -- 1 07 .

Si lva-Corvalan. Carmen 1 99 1 Spanish language attrit ion i n a contact situation with English. I n : Seliger and Vago (eds . ) ,

1 5 1 - l 7 l . Si lva-Corval,\n , Carmen

1 994a Language Contact and Change. Oxford: Clarendon Press. Si lva-Corval{lIl, Carmen

1 994b The gradual loss o[ mood distinctions in Los Angeles Spanish. Language Varimion and Change 6( 3 ) : 255- 272.

St-Amand, Anne 2002 Le subjonctif suivant une expression non-verba Ie. Unpublished M A thesis (memoire).

Department of Linguistics, University of Ottawa. St-Amand , Anne and Shana Poplack

2002 A real-t ime window on 1 9th century vernacular French: The Rh'its du ji·an(,ai.\· ljlll!iJecois d'autrejois. Paper presented at the annual meeting of the Canadian L inguistics Associat ion, University of Toronto, Toronto.

Tagl iamonte, Sal i 2002 Comparative sociolinguistics. I n : Jack Chambers, Peter Trudgil l and Natalie Schi l l ing

Estes (eds. ) . Handbook (Jf Language VClriation and Change, 729- 763. Malden/Oxford: B lackwel l .

Thomason, Sarah G. 200 1 Language Contact: An Introductiol1. Washington, DC: Georgetown University Press.

Thomason, Sarah G. 2007 Language contact and deliberate change. Journal oj' Language Contact I : 4 1 - 62.

Thomason, Sarah G. and Terrence K au[man 1 988 Language Contact, Creoli::ation, and G('I7etic Linguistics. Berkeley/Los Angeles: Univer

sity of California Press. Torres-Cacoul los, Rena and Catherine E. Travis

to appear Variable )'o expression in New Mexico: English influence? I n : Susana V. R ivera-Mi l l s and Daniel J . Vi l la (cds.), Spanish o/ the Southwest: A Language in Transition . Madrid : I beroamerican a/Vervuerl.

Toribio, Almeida Jacqueline 2004 Convergence as an optimization strategy in bil ingual speech: Evidence from cocleswitch

i ng. Bilingualism: Language and Cognition, 7(2): 1 65 - 1 73 . Van Coetsem, Frans

2000 A General and Unified Theor), of' the Transmission Process in Language Contact . Heidelberg: Winter.

Weimcich, Uriel, William Labov and Marvin Herzog 1 968 E mpirical foundations [or a theory of language change. I n : Winfred P. Lehmann and

Yakov Malkiel (cds.) . Directiol1s jbr Historical Linguistics. 97- 1 95. Austin : U niversity of Texas Press.

Whitney. Will iam Dwight 1 88 1 On mixture i n language. Trwlsactio/lS of'the American Philosophiml Association 1 2 : 1 - 26.

Will is. Lauren 2000 Etre Oli nI' plus ('tre: Auxil iary a lternation in Ottawa-Hul l French . U npublished MA thesis

(memoire). Department of Linguistics, University of Ottawa. Winford. Donald

2003 An Introduction to Contact Linguistics. Oxford: Blackwel l .


23 . Areal language typology

Winford. Donald 2007 Some issues in the study of language contact. Joumal oj' Languag(' ContaCl - Th!'m({

(2007) . Available online from <http://www. jlc-j ournal.org > . Zentz. Lauren

2006 "ecst t;a je travaille dessus ": Orphaned prepositions and relativization in Canadian French. Unpublished MA thesis (memoir!'). Department of Linguistics. U niversity of Ottawa.

Zribi-I-1ertz, Anne 1 984 Orphan Prepositions in French and th(' Concept oj' "Nu/! Prol10lln " . Indiana University

Linguistics Club: Bloomington, I N . Zwicky, A .

2005 More i l l usions. Available at < http://i trc.cis.upenn .edu/-mylllanguagelog/archives/002407. html > . Accessed 29 April 2008.

S/zan({ Poplack and Stephen Levey, Ott([1I'({ (Canada)

23 . A rea l l a n guage typo logy

I . I ntroduction: Setting the stage 2. Language typology 3. Language change: The social and linguistic dynamics of contact-induced convergence

and areal i ty 4. Consequences for Linguistic Typology and i ts generalizations 5 . Conclusion and outlook 6. References

1. Introduction: Setting the stage

Linguistic typology deals with l inguistic variation and the discovery of cross-linguistic universal patterns reflected in that variation. The aim of the present article is to show that the robustness and the reliability of typological generalizations crucially depends on the appropriate assessment of areality, i. e., the inl1uence of geographic closeness and language contact on structural properties of languages. Thus, the analysis of structures and their overall statistical distribution i n the popUlation of the world's languages does not guarantee any direct access to valid generalizations about language universals and about properties of the human cognition.

This can be illustrated by an example f rom basic word order. In his 1986 volume, Tomlin looked at the overall percentage of the six logically possible combinations of S ( Subject), 0 (Object) and V ( Verb) in a statistically balanced sample of 402 languages. As it turned out, the majority of these languages were either SOY (180 languages, 44.78 percent) or SVO (168 languages. 41.79 percent). Only 37 languages (9.20 percent) were VSO. The sequences in which 0 preceded S were only represented by a small minority of twelve languages 12.99 Dercent) with VOS. five languages in the Amazon ian re 2'i on o f

4 1 9


05 02-33-27PoplackLeveyInPress

Documents

Transcript of 05 02-33-27PoplackLeveyInPress